Deepot: Parking Lot Identification Using Low-Resolution Satellite Imagery
Files
Date
2025-01-07
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Narrator
Transcriber
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
Volume
Number/Issue
Starting Page
5315
Ending Page
Alternative Title
Abstract
Detecting parking lots is essential for siting electric truck charging infrastructure, ensuring convenient access for drivers. However, the private ownership of many parking lots limits public access to information for planners. Current object detection methods rely heavily on high-resolution satellite imagery, which is often restricted in availability. To address this, we introduce Deepot, a deep learning-based truck parking location identification approach using low-resolution satellite imagery. We begin by retrieving geographical data for the target area and optimizing image resolution for model training. The model produces initial parking lot locations, which are then converted to real global latitude and longitude coordinates using a custom coordinate reference systems transformer. These global coordinates facilitate querying Google Maps for high-resolution images to enhance detection performance. Deepot streamlines charging infrastructure planning from diverse satellite imagery and offers additional, high-fidelity candidate locations to medium- and heavy-duty infrastructure planning tool, HEVI-LOAD. Extensive experiments show the effectiveness of Deepot.
Description
Keywords
Location Intelligence Research in System Sciences, deep learning, electric truck charging infrastructure, low-resolution satellite imagery, object detection, parking lot detection
Citation
Extent
10
Format
Geographic Location
Time Period
Related To
Proceedings of the 58th Hawaii International Conference on System Sciences
Related To (URI)
Table of Contents
Rights
Attribution-NonCommercial-NoDerivatives 4.0 International
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.